Avri wrote, Run of the RFC3797 program Type size of pool: (or 'exit' to exit) 2 Type number of items to be selected: (or 'exit' to exit) 1 Approximately 1.1 bits of entropy needed. Type #1 randomness or 'end' followed by new line. Up to 16 integers or the word 'float' followed by up to 16 x.y format reals. 2 20 21 38 50 2 20 21 38 50 Type #2 randomness or 'end' followed by new line. Up to 16 integers or the word 'float' followed by up to 16 x.y format reals. end Key is: 2.20.21.38.50./ index hex value of MD5 div selected 1 EFC9E973E94E5D3BB70D8CDE6C0EDBCD 2 -> 2 <- ------------------------------------------------ It seems to me that there is a programing bias here. Entropy (see above) can be defined at the state of disorder in the universe or more generally as a doctrine of inevitable social decline and degeneration. The entropy of the universe is inceasing. Since ICANN's inception, our activities have dramactically contributed to the state of disorder of the universe (witness the GA list or an ICANN workshop) and almost certainly to degeneration. Hence we have ourselves introduced a bias in the "entropic need" specified above. This selection thus cannot be random.